By Ben Zimmerman, Feeding Clinic of Los Angeles
Michael was a 4-year-old child diagnosed with autism.
He had an unremarkable medical history. Prior to coming to the clinic he would only consume bottles of formula. He would drink 6-8 bottles per day of concentrated formula. He would not drink from other containers nor would he drink other liquids. He also did not consume any solids.
He was seen for four weeks in total. Treatment was three times per day, five days per week.
The first week consisted of introducing solids to him. Purees were used for two reasons: (1) he had never chewed on a solid, which could make sessions dangerous because poorly chewed solids can be a choking hazarding (2) more trials can occur per session because time is not wasted on chewing.
Four foods, chicken, green beans, macaroni and cheese, and applesauce, were randomized throughout the meals and week.
Introducing solids consisted of presenting a bite, while simultaneously verbally prompting him to take a bite. The bite was presented to the top lip until there was a mouth opening, and then the bite was deposited in the mouth. The bite would not be deposited in cases of gagging, coughing or vomiting, but were put in during all other openings. Each meal had a time cap of twenty-five minutes, and volume of seven ounces of food. The session would end when either the time cap had elapsed or the volume was consumed, whichever occurred first.
The chart above shows how the time between presentation of food and the bite being taken decreased. During the first session bites were taken an average of ten seconds after food presentation. By session three, bites were taken on average after three seconds, and by session five they were stable at about one second.
Week two of treatment involved the introduction of new foods. Bites were presented with the same protocol as in week one. Four novel foods were presented in each session. Volumes of each food were held constant at three ounces each, including all food groups in the meal. Randomizing foods in this way, instead of focusing on one or just a few foods helped to generalize eating across all foods, as can be seen in the next graph.
There was a general trend in latencies dropping over time. By the tenth session of week two, latencies were stable at or near one second. This graph shows that eating was conditioned to occur at short latencies across all foods.
Generalizing meals to other settings serves the function of not only strengthening eating behaviors, but also ensuring that eating would take place across all settings (no matter the food), including the child’s own home environment. Week three consisted of randomizing nine settings across the fifteen meals throughout the week. All other variables were held constant, including randomizing the same foods used in week two. Settings included the park, playroom, different treatment rooms, hotel room, and hotel lobby.
There was an initial spike of refusal, which was followed by a sharp downward trend. By session 35 latencies were at or near one second.
Week four consisted of training primary caregiver to feed with random foods and random settings. This would ensure that feedings across settings and foods would only take place in the context of the therapist, but also ensure that feedings take place with the primary caregiver.
There was a downward trend after session 47. Most meals thereafter were at one or two seconds, except for a spike in latency for sessions 52-53. Primary caregiver reported that eating continues at low latencies across foods and settings. This was reported four months subsequent to completion of treatment.
Data was collected by videotaping all sessions. Two data collectors independently took data while viewing video footage at separate times. Inter-observer reliability was measured at 86%.